Institutions, policies, and arguments: context and strategy in EU policy framing

Institutions, policies, and arguments: context and strategy in EU policy framing Paper to be presented at the section Interest Groups and Comparative Political Science: Empirical, Theoretical, Methodological and Normative Challenges, Panel: Interest Groups and Social Movements at the ECPR General Conference, Glasgow, September 3 th 6 th 2014. Rainer Eising, Daniel Rasch, Patrycja Rozbicka, Ruhr-University Bochum Abstract Studies of framing have shown that policy actors highlight certain aspects of policy proposals to impact on legislative outcomes. However, little is known about the determinants of framing activities in EU public policymaking. In this article, we are mostly interested in the interplay of institutional and policy contexts with the strategic side of framing activities. We draw on framing theory, policy analysis and a conception of the European Union as a multilevel system to develop our arguments about the impact of contexts and strategy on the framing dynamics in EU policymaking. We use a computer assisted manual content analysis to identify the frames that were prevalent in four policy debates at the EU level and in Germany, Sweden, the Netherlands and the United Kingdom. The results demonstrate that, besides the EU s institutional and policy contexts, strategic highlighting of specific dimensions of EU policy proposals has also effect on framing activities. Key words: framing, context, strategic action, interest groups, EU multilevel political system 1

1. Introduction Frames highlight specific aspects or dimensions of a policy problem while omitting others. They can play a crucial independent role in public policy debates and impact on their outcomes (Baumgartner and Jones 1993). It is therefore surprising, that we know rather little about the emergence and variation of policy frames in the European Union, especially with regard to contextual factors. To enrich the debate about how European policies are created and if policy arguments make a difference to policy outcomes, we address three questions about framing processes in EU policymaking: First, what policy frames emerge in EU policy debates? Secondly, can we explain the type and the number of frames (which we term framing activities) based on contextual factors, like national or EU level conditions under which they came to existence (institutional context), and the differences between policy areas in which they were employed (policy context)? Thirdly, alternatively to context or in conjunction with it, can strategic highlighting of policy dimensions explain framing activities? We draw on framing studies, policy analysis, and conceptualisations of the EU institutional setting as a multilevel system that engages mostly in regulatory policymaking to develop our arguments about the role of institutional and policy contexts in which EU policies are enacted. Conceptualising the European Union as multilevel system generates the expectation that actors at different levels will engage in framing activities and that the actors frames will differ depending on the level on which actors are situated as well as across national sub-units in this system (see Klüver, Braun and Beyers, 2014). Conceiving of the EU as a polity that is mostly engaged in regulatory policymaking yields the expectation that, regardless of the policy area and policy proposal in question, frames will refer to various aspects of EU 2

regulation and harmonization. However, drawing on policy studies leads also to the expectation that we will identify frames that matter across policy areas such as frames about the administrative and economic burdens of a policy proposal, and frames that relate specifically to environmental policies, to financial market policies, to education policies, etc (see also: Klüver, Mahoney and Opper, 2014). In addition, we draw on framing studies which add to the debate that policy framing is a strategic activity that highlights specific dimensions of a policy proposal. This argument, in contrast to explaining framing activities by use of context, focuses on deliberate action of the involved actors and their strategic use of frames. We expect that a greater variety of dimensions in a policy proposal leads to a greater variety of frames in the policy debate. In our study, we identify the policy frames that were prevalent in four EU policy debates which were initiated between 2008 and 2010. We focus on two environmental policy debates and two instances of financial market regulation. Our research design includes variation across the EU and the national levels, across four member states (Germany, the Netherlands, Sweden, and the United Kingdom), as well as across and within policy areas. To study the policy frames, we use a computer assisted manual content analysis (for details see: Boräng et al 2014) that was conducted on a sample of the position papers submitted to EU level or national level consultations, as well as the official documents of the political institutions. The paper proceeds as follows. In the next section, we illustrate our theoretical framework in detail and develop the hypotheses that guide the empirical analysis. Afterwards, we explain the research design of our study, followed by a presentation of the empirical analysis. First, we highlight which frames emerged in the four policy debates and demonstrate their correspondence with the EU s institutional characteristics and the two policy areas. Secondly, we develop a statistical model that seeks to account for the extent of framing activities. We 3

conclude with a summary of our results and a discussion of the broader implications of our findings. 2. The emergence and variation of frames in the EU multilevel system Framing theory suggests that since every policy issue has multiple potential dimensions, framing selecting and emphasizing particular aspects of an issue is an important tool for policy actors (Daviter 2009; Baumgartner and Mahoney 2008). According to Entman frames highlight some bits of information about an item that is the subject of communication, thereby elevating them in salience (Entman 1993: 53). We consequently define framing as selecting and highlighting some features of reality while omitting others (Entman 1993: 53). The selected and highlighted features of the policy proposals denote what actors perceive to be at stake in an issue (Daviter 2009: 1118). And what is perceived to be at stake is likely to matter greatly for the scope of a political conflict which, in turn, will have a large impact on its outcome (Schattschneider 1960). In this article, we are mostly interested in the interplay of institutional and policy contexts with the strategic side of framing activities in the EU. The micro-foundation of our analysis consists of behavioural assumptions about policy advocates in EU policymaking. First, actors pursue their self-interests. While there are manifold motives for organizational action (Mansbridge 1990), the actions of policy advocates such as interest groups are usually meant to promote the rights or the material benefit of their members or constituency (Olson 1965). Secondly, these organizations act rationally, but their rationality is bounded. Choosing their actions from a set of alternatives that are ranked according to a transitive preference ordering, they evaluate their alternative options according to a general rule and engage in framing activities to promote their interests. However, they do not necessarily possess full information about the decisional situation, face computational limits in processing their information and 4

can therefore not always make the best possible choice. Because of their bounded rationality, they engage in behaviour that is conditioned by the context in which they are embedded. 2.1 Institutional context the EU as a multilevel governance system After more than sixty years of European integration, the EU political system and EU politics are highly institutionalised. There exists a widely shared system of rules and procedures to define who the actors are, how they make sense of each other s actions, and what types of actions are possible (Stone Sweet et al. 2001: 12). An important element of this institutional setting is the strong focus on the creation, maintenance, and policing of the internal market through regulatory policies (Majone 1996). Departing from the assumption that the EU institutional context determines, to some extent, which frames emerge in EU legislation we should expect that policy advocates focus an important part of their framing efforts on EU market integration, EU regulation and EU policy harmonization. While the relevance of such institutional frames may vary across policy proposals, they should nonetheless matter in most EU policy areas. Our understanding of the European Union as a multilevel system in which EU level and member state institutions share policymaking authority in many areas and in which the member states have important decision making powers at EU level, draws attention to the manifold relations that have developed between sub-national actors, interest groups, and national institutions on the one hand, and the EU institutions on the other (Marks 1993: 402). At the European level, this perspective emphasizes the need for cooperation and coordination between involved stakeholders and, in particular, a crucial role for the European Commission as a policy entrepreneur who takes skilled action to construct or revise policy frames which will bring together the relevant political and societal actors (Stone Sweet et al. 2001; Daviter 2009). Since she is highly dependent on external information, we do not focus our attention 5

exclusively on the European Commission, but depart from the assumption that a variety of policy advocates is engaged in a framing competition to shape EU legislation. Furthermore, national governments are key to EU decision making and act partly as gate-keepers between domestic and European affairs. While many national interest groups have joined EU level groups and engage also in direct lobbying activities vis-à-vis the EU institutions, for a large number of them, the national governments still remain the most important contact partners when representing their interests in EU legislative processes (see also: Eising, 2004). The member state governments policy preferences are to a large degree shaped in discussions and negotiations with domestic interest organizations (Moravcsik 1998). Accordingly, the frames of these national actors depend on the national economic and societal interests and the degree to which already existing national regulation would have to be adjusted due to new EU regulation (Green Cowles et al. 2001: 12-14). In sum, we posit that: Hypothesis 1.1: The framing activities of actors located at the EU level differ from those located at the national level. Despite their embeddedness in the European integration process, major institutional characteristics of the EU member states continue to vary in important ways: their varieties of capitalism in terms of coordinated and liberal market economies (Hall and Soskice 2001), their welfare states in the form of liberal, corporatist, and social-democratic welfare regimes (Esping-Andersen 1990) and their patterns of interest mediation in the corporatist and pluralistic modes (Schmitter 1974). Previous studies have demonstrated that such differences shape the actors policy preferences (see Ringe 2005). It follows that the national borders of the EU member states promote a rivalry over problem views and solutions which have to be accommodated in EU policymaking. Consequently, we should find significant cross-national variation in the types of frames and the extent of framing activities in EU policy debates: Hypothesis 1.2 Framing activities vary across EU member states. 6

2.2 Policy context A variety of studies have pointed out that policy areas differ in terms of the outputs they typically produce. They generate regulatory, distributive, and redistributive policies in varying mixtures, and such variations can lead to differences in policymaking processes. Furthermore, the institutional settings and the allocation of competencies to political institutions can differ systematically across policy areas. This shows not only in the distinction of the EU s exclusive, mixed, and coordinating powers but also in the variation of governance modes and actor constellations across policy areas (see Wallace 2010). The underlying problem structures and ideational foundations of public policies in different policy areas differ also significantly. The high degree of vertical and functional segmentation in the EU institutions is likely to perpetuate such variations because the actors in the different policy sub-systems develop their own rationality criteria over time. We thus expect that: Hypothesis 2.1: Framing activities vary across EU policy areas. 2.3 Strategic action The study of framing activities adds to the debate that actors may strategically highlight specific dimensions of public policies. Paying attention to their strategic choices allows us to disentangle the strategic element in framing patterns from the contextual factors that impact on them. Assuming that policy framing is a rational strategy and that each actor is specifically located within the institutional and policy context yields the expectation that the number of frames tends to increase with the number of actors that become vocal on the policy proposal. This proposition modifies Schattschneider s expectations (1960: 16-17) about the expansion of conflicts in a democratic polity. He identifies three factors as being crucial to an expansion of conflicts in the sense of bringing in more actors, namely the visibility of the conflict to the public, the amplitude of the powers and resources of the political institutions that respond to the conflict situation, and the policy advocates competition about the expansion of the scope 7

of conflict. We suggest that the first two factors are limited in the European Union. The lack of a European public is often bemoaned, and while the EU powers have been expanding over time, they are still restricted. The third factor, political competition, is present but in a modified form: Against Schattschneider s focus on political competition bringing in larger publics into EU legislative processes, we suggest that it is largely aimed at shifting the debate by highlighting a different aspect of the proposal and putting into question the scope, the validity, or the legitimacy of others arguments. Consequently: Hypothesis 3.1: The more actors are involved in a policy debate, the greater are the framing activities. Furthermore, we expect that the framing activities depend on the stakeholders positions on the policy proposals. Based on Schattschneider s proposition (1960: 16-17) that it is the losers of political contests who seek to expand the scope of conflicts, we expect that the framing strategies of actors who oppose a proposal, because they expect to suffer losses through it, differ from those of actors who are in favour of a proposal or are undecided about it. This leads to the hypothesis: Hypothesis 3.2: The framing activities of actors who oppose a proposal differ from those of actors who are in favour of it and from those who are undecided or neutral about it. To identify the dimensions of policy proposals the involved actors are likely to highlight, we build on Entman s (1993) definition of frames. He suggests that policy advocates highlight at least the four following dimensions of policy proposals: problem definition, causal diagnosis, moral evaluation, and ability to remedy the underlying problem. First, in terms of the definition of the underlying problem, the interests of policy constituencies give rise to interest-based arguments about the nature of the problem. Secondly, with regard to the causal analysis of the underlying problem, policy advocates will put forward arguments about the relevance of the policy proposal, thus contributing to the framing activities in a policy debate. 8

Thirdly, in terms of moral judgements, framing activities increase if policy proposals are based on normative justifications or have normative implications that may be supported or contested. Fourthly, calculations about the likely impact of the proposed policy measures in terms of their social welfare effects or their supposed financial, administrative, or economic burdens address the remedial capacity of the proposal. This leads to the following propositions: Hypothesis 3.3: The dimension of constituency interests enhances framing activities. Hypothesis 3.4: The causal relevance of EU policy proposals will result in an increase of framing activities. Hypothesis 3.5: The normative dimension of EU policy proposals stimulates an increase of framing activities. Hypothesis 3.6: The remedial capacity and cost-benefit dimension of EU policy proposals will result in increased framing activities. 3. Research design We test these hypotheses in a comparative study of four EU policy debates at the EU level and in four EU member states, namely Germany, the Netherlands, Sweden, and the United Kingdom. This selection assures variation with respect to country size, duration of EU membership, varieties of welfare state and capitalism, and the state-interest group model (see Hall and Soskice 2001; Lijphart 1999; Esping-Andersen 1990). Given that earlier studies identified a substantial impact of such structural differences on policy preferences in EU politics (see Ringe 2005), we expect to observe different types of frames and levels of framing activity within the four countries. To gauge the relevance of policy areas for framing activities, we analyse two policy proposals in the area of financial market policy and two proposals in the area of environmental policy. While economic policies and especially the internal market have a long lasting history of EU wide regulation, the field of financial market regulation is a relatively recent area of EU 9

market integration and regulation. The current tendency towards a more detailed and comprehensive regulation of financial markets meets, in part, with strong national opposition. In comparison, preceded by legislation that was designed as a by-product of internal market EU developments, EU environmental policy started in the 1980s with strict measures. Today, environmental policies focus more on setting substantial goals rather than on prescribing strict policy instruments thus allowing for more flexibility for the addressees. Given such differences between the two policy areas, the expectation is that framing activities in EU financial market regulation and in EU environmental policy will differ. We selected the four policy proposals in our study based on the following criteria. We took all European Commission Directives that were introduced between 2008 and 2010. We then identified those 20 policy proposals which surpassed a certain level of salience in different print media at the national and at the EU level: The proposal had to be mentioned in at least one European level media source (Agence Europe, European Voice)and two national level outlets (Financial Times, Le Monde, or Frankfurter Allgemeine Zeitung). Setting a minimum level of salience was meant to ensure that there were policy debates about these proposals. In the last step, to investigate whether different proposal characteristics lead to different framing activities in one and the same policy area we took the two most salient proposals from each of the above mentioned policy areas. Consequently, we focus in this paper on two financial market directive proposals that are part of the effort to strengthen financial market stability following the global financial crisis: the directive proposals on alternative investment fund managers (AIFM COM 2009) and deposit guarantee schemes (DGS COM 2010). The former aimed at harmonizing the requirements for entities engaged in the management and administration of alternative investment funds. It was proposed in June 2009 and passed the EU s legislature one year later. The latter was introduced in 2010 to revise earlier legislation on this subject. It is meant to 10

protect savers and to prevent bank runs in case of a bank s bankruptcy as well as to harmonize national deposit protection schemes. The European Parliament passed the directive in April 2014, based on an agreement that was reached in the trilogue with the Commission and the EU Council in December 2013. The two energy/environmental cases focus: on the disposal of electrical and electronic waste (WEEE - COM 2008a) and on the promotion of the use of renewable energy resources (RED - COM 2008b). The RED directive proposal aimed at establishing an overall binding target of 20% share of renewable energy sources in energy consumption and of 10% for the use of biofuels in transport to be achieved by each member state by 2020. The directive was proposed in January 2008 and passed the EU s legislative process in June 2012. The WEEE proposal was introduced to revise an earlier directive from 2003. It mostly addressed the administrative problems and costs caused by the original directive. The recast proposal was tabled in December 2008 and was passed in July 2012. To identify policy frames, we rely on a computer assisted qualitative content analysis. The main advantage of the method is that it can be applied to documents in different languages and allows the study not only of the manifest meanings and frequencies of words but also of their latent and contextual meanings (e.g. Schreier 2012). To study the policy frames we developed a codebook that was implemented in the software MAXQDA. The codebook drew on insights of previous framing studies in the fields of communication studies, psychology, and social sciences and has been designed with the unidimensionality, mutual exclusiveness, and exhaustiveness of the codes in mind (Schreier 2012: 71-77). The validation of the content analysis is based on cross-checks with the statements of interest groups in expert interviews. Krippendorff's alpha for inter-coder reliability was 0.739, with an 87.44% agreement. The calculations include the not-coded segments as agreements (when coders agreed to not assign the code). 11

Our understanding of frames reconstructs the actor s perception of the essence of the directive proposal. To identify these essential frames, the coders marked those passages in the selected texts which indicated: What the author of the text thinks is at stake in this proposal at a more general level?. After identifying these passages, we defined different types of frames based on the empirical findings and summed up the number of frames in each document. For the first distinction of our typology, we drew on de Vreese (2005) who distinguished among generic and issue specific frames. Whereas issue-specific frames are tied to the political issue at hand, generic frames can be applied across policy areas (de Vreese, 2005). In order to relate the types of frames to institutional and policy contexts, we developed his classification further by distinguishing among institutional frames that relate to the general rules of the EU political system and policy frames that relate to substantial policy goals, norms, and instruments, thus leading to a fourfold classification scheme. 4. Empirical analysis The empirical analysis proceeds in two steps. First, we describe the types of frames that have emerged in the four policy debates and elaborate them in the different contexts that we suggested in earlier sections. Secondly, we seek to explain the extent of framing activities in terms of the type and number of frames that have emerged in these debates. 4.1 Descriptive evidence Using the EU level and the national governments and parliaments databases, media and web mining, we identified 1,093 public and private stakeholders that were involved in the four policy debates. Not all stakeholders submitted position papers, and we could not trace all submitted position papers. In total, we collected 704 documents from 443 actors. The majority of the statements were submitted in consultation processes organized by the European 12

Commission or the national authorities. We selected a stratified sample of documents based on the number of available documents on each proposal and the types of actors involved in the policy proposals. The procedure resulted in a sample of 280 documents written by 309 actors, among them 73 German actors, 74 Dutch actors, 37 Swedish actors, 61 British actors, and 64 EU level actors. 268 actors presented one policy document, and 41 actors tabled 15 joint position papers. About 79% of the documents are position papers drawn up by interest groups, think tanks and companies, the rest being government studies, expert reports, etc. Based on the findings of our content analysis, 40% of the policy actors did not mention an essential frame while 60%, or 186 of the 309 policy actors, referred to the essence of the proposal. 36 of these actors identified three or more frames, 47 referred to two frames, and 103 mentioned just one frame. The bulk of the actors mentioning more than two frames were involved in the two financial market proposals: 27 actors relative to 9 such actors in EU environmental policy. Table 1 indicates the distribution of the frames across the four policy debates. We identified 14 essential frames in the four policy debates. In the debate on the WEEE directive, seven frames mattered, in the debates on the RED and the DGS directives we identified eight frames, and in the debate on the AIFM directive nine frames played a role. Accordingly, EU policy debates are, in part, framing competitions. As suggested, generic institutional frames shape policy debates across various EU policy areas, underlining that the EU s rule system weighs heavily on the types of frames that typically emerge in EU policy debates. These frames relate to EU regulation and harmonization as well as the administrative (and economic) burdens attributed to the EU policies. Furthermore, the frames on the implementation of EU policies and information and transparency were important in both policy areas even though not across all four policy proposals. As a highly specific institutional frame and regulatory tool, the setting of mandatory targets was a major frame only in the renewable energies proposal. 13

Policy frames were more confined in their scope than the institutional frames. Among the supposed generic policy frames, only market integration was highlighted in both policy areas. The health, consumer safety, and employment frames were debated in just one policy area and in just one or two (consumer safety frame) policy debates, underlining that there are substantial differences among the four EU policy proposals. Specific policy frames could be identified in both policy areas: Financial market stability played an important role in the debates on both the AIFM and the DGS directive proposals. An environment frame guided the debates on both the WEEE and the RED directive proposals while the climate and the energy frame mattered only in the latter. In sum, institutional frames tend to contribute to a convergence of frames across policy areas, while, apart from the market integration frame, policy frames tend to contribute to the variance of frames across policy areas and proposals. TABLE 1 ABOUT HERE 4.2 Accounting for the number and types of frames in EU policy debates In the next step, we account for the extent of framing activities and the types of institutional and policy frames that were invoked in the four policy debates. Our explanatory variables are derived from the coded documents of the stakeholders in the policy debates. They cover the location of the actors in the EU multilevel system, important characteristics of the policy context, and the strategic mobilization of actors and their highlighting of policy dimensions. They have been coded independently from the essential frame variable and are described in the table 2. In the models below, we also control for the extent of change a policy proposal is perceived to bring about (routine change or major change), whether a policy is sectoral or cross-sectoral in its scope, and the type of actor who voiced the frame (public interest group, business interest group, state actors, institution, or firm). TABLE 2 ABOUT HERE 14

We analyse the impact of the experimental variables by means of Poisson regressions. We calculated four models because we have four dependent variables: (1) the total number of frames invoked by an actor, (2) the number of generic institutional frames, (3) the number of generic policy frames, and (4) the number of specific policy frames invoked by an actor. 1 These are count variables that are not overly dispersed. The top panel in table 3 presents the regression results. Raw coefficients are Poisson regression coefficients. The significance levels are based on two-tailed tests. The bottom panel presents the summary statistics of the maximum likelihood estimations. All models are statistically significant, and it appears that the model explaining the extent of generic policy frames works better than the models accounting for the overall number of frames and the other types of frames. In general, taking into account the framing strategies of actors in EU policy debates on top of institutional and policy context helps accounting for EU framing dynamics. TABLE 3 ABOUT HERE The evidence presented in table 3 indicates that the framing activities of EU level actors and national actors differ significantly in the EU multilevel system supporting the hypothesis 1.1. EU level actors tend to employ a smaller number of frames than Swedish actors, the reference category, when arguing for or against policy change. While they do not resort to generic institutional and policy frames in any other way than national actors, they tend to employ fewer specific policy frames than these. Being located at the apex of national associational systems, EU level interest groups seek to bridge national heterogeneity and aggregate the frames of their members in their position papers. Their interest aggregation function leads them to highlight fewer and more general frames than the national actors who voice their 1 We did not include the specific institutional frame mandatory targets as a dependent variable in these analyses because it was invoked by a very small number of actors and is a binary variable. 15

positions based on domestic structures and interactions in EU policymaking. In fact, 64% of the EU level actors do not identify any policy frame whereas this applies only to a third of the national associations. There are also significant cross-national differences in the EU multilevel system. The British and German policy actors invoke fewer frames than Swedish actors which supports hypothesis 1.2. Note that British and German actors invoke even fewer frames than the EU level actors. However, national actors do not differ significantly in their usage of different types of frames when controlling for contextual factors and the strategic highlighting of policy dimensions. Accordingly, there are significant differences in the overall level of framing activities among the actors located in different countries but these differences are not systematically related to the types of frames (institutional or policy frames) they invoke. Hence, the institutional differences of the member states prove influential for the level of framing activities but the substance of these activities seems to be shaped to a greater extent by the EU policy characteristics. In line with this expectation, the policy context proves influential when it comes to framing activities in the EU multilevel system. The general characteristics of policy areas impact decisively on the framing activities. In our cases, the actors involved in EU environmental policymaking invoke fewer frames than those in EU financial market regulation supporting hypothesis 2.1. In particular, the policy advocates in EU environmental policy tend to invoke fewer policy specific frames than those involved in financial market integration. To some extent, the discrepancy may be attributed to different policy origins. The reregulation of financial markets has a major exogenous policy origin: the global financial crisis has opened up new courses of action in the policy menu and put into question established policy paradigms. EU environmental policy has more gradually evolved since the 1980s and not been subject to such a paradigm change at the end of the past decade. 16

Furthermore, table 3 illustrates that the strategic decisions of actors leave an imprint on the framing activities in EU policy debates. First, the extent of political mobilization impacts on the nature of the political debate which supports hypothesis 3.1. The more actors decide to engage in a policy debate, the more frames are invoked. Political mobilization leaves also a mark on the nature of the debate: The more actors are involved, the more policy oriented the debate becomes, both in terms of generic and specific policy frames. Correspondingly, the denser the population of actors in a policy debate, the less recourse to generic and specific aspects of the EU rule system do the actors have. Secondly, other than expected in hypothesis 3.2, the policy positions of actors do not impact on the number or types of policy frames in the debates. Voicing opposition or support to an EU policy proposal has no effect on the number of frames actors invoke or on their usage of institutional or policy frames. Note that this finding does not rule out that the positions that actors represent in EU policymaking may be associated with particular frames. In our cases, actors who opposed a Commission proposal were more inclined to refer to the regulation or the consumer safety frames than actors in support of the Commission proposals. And actors in support of a Commission proposal were more likely to refer to the environment or the mandatory targets frames. Finally, our hypotheses about the impact of the highlighting of policy dimensions on the extent of framing activities find general support (Hypotheses 3.3, 3.4, 3.5, 3.6). Framing theory suggests that actors employ frames in order to highlight specific aspects of the policy proposals. Accordingly, the strategic arguments actors make about the different dimensions of a policy proposal tend to raise the number of essential frames in the policy debate. This holds for arguments which highlight the relevance of policy proposals as much as for arguments about constituency interests and claims about the normative implications and the likely costs and benefits of a proposal. But these arguments bear differently on the types of frames invoked in the EU policy debates. In other words, these four policy dimensions are 17

associated in different ways with institutional and policy frames. Highlighting the administrative and economic burdens EU policies may impose is a highly proposal-specific framing strategy as it is associated only with an increase in the number of policy specific frames. Actors who make these arguments do not differ from other actors in their usage of institutional frames or generic policy frames. In contrast, actors making normative claims relate as much to specific policy principles such as the precautionary principle in EU environmental policy as to generic institutional norms of European integration. Furthermore, highlighting the relevance of policy proposals is associated with the generic institutional frames and the EU s focus on political regulation and policy harmonization rather than with presentation of policy frames. Emphasising constituency interests has the opposite effect : It is associated with an increase of policy frames, be they generic or specific, but is not associated with the usage of generic institutional frames. In sum, only one of the four dimensions is associated with both policy and institutional frames (normative evaluation). Highlighting one of the other dimensions increases the density of either institutional frames (policy relevance) or the policy frames (constituency interests, costs and benefits) Policy advocates highlight various policy dimensions in the form of frames to impact on the outcome of EU legislation. However, developing frames in order to expand the scope of conflict in EU policymaking comes at a cost. The empirical analysis suggests that other actors will engage in counter framing activities leading to a greater density of mostly policy specific frames in EU legislative debates. 5. Conclusions This article has studied the effect of contextual factors and of the strategic highlighting of policy dimensions on the types and number of frames that have emerged in four EU policy 18

debates. The results of our study indicate that the EU s institutional context exerts a significant impact on the essential frames of the policy advocates in the EU policy debates. The location of actors at specific levels and in specific member states causes variations in their framing activities. The institutional context also promotes the emergence of generic institutional frames about EU regulation or harmonization in different policy areas. The EU s emphasis on market integration furthers the emergence of generic policy frames that are associated with market making and market competition. The policy context has also great repercussions on the framing activities. It promotes differences across policy areas such as financial market policy and environmental policy, and in part also differences across proposals within policy areas. The article highlighted that the strategic element in framing activities must not be neglected. Policy advocates are not forced by the institutional and policy context to employ specific frames; they can choose which dimensions of a policy proposal to highlight. Specific dimensions of proposals such as their problem solving capacity provide incentives for argumentation strategies that are associated with an increase of essential frames. Highlighting the normative dimension of EU policy proposals, their effect on constituency interests, or their problem-solving capacity typically increases the density of specific policy frames in EU policy debates. The impact of these argumentation strategies on generic institutional or policy frames varies to a greater extent with the type of argument put forward. The more actors become involved in the policy debates, the more the frame competition intensifies. Any effort to expand the scope of conflict in EU policy debates through framing is likely to lead to counter framing and to a greater variety of specific policy frames so that the outcome of such a conflict expansion is rather uncertain. It is difficult for individual actors to win the day with their frames, such that a mix of frames could be identified in each of the four 19

EU policy debates. As a corollary, the importance of the frames varies across policy proposals. Word count: 6772 (text and references) + 1425 (tables, 3*475) = 8197 20

References Baumgartner, F. R. and Jones, B. D. (1993) Agendas and Instability in American Politics, Chicago: University of Chicago Press. Baumgartner, F. R., and Mahoney, C. (2008) The two faces of framing Individual-level framing and collective issue definition in the European Union, European Union Politics 9(3): 435-449. Boräng, F., Eising, R., Klüver, H., Mahoney, Ch., Naurin, D., Rasch, D, Rozbicka, P. (2014) Measuring frames: A comparison of methodological alternatives, Interest Groups and Advocacy 3 ( 2): 188-201 COM (2008a) Proposal for a Directive of the European Parliament and of the Council on waste electrical and electronic equipment (WEEE) (Recast), COM (2008) 810. COM (2008b) Proposal for a Directive of the European Parliament and of the Council on the promotion of the use of energy from renewable sources, COM (2008) 019. COM (2009) Proposal for a Directive of the European Parliament and of the Council on Alternative Investment Fund Managers and amending Directives 2004/39/EC and 2009/ /EC. COM (2009) 207. COM (2010) Proposal for a Directive / /EU of the European Parliament and of the Council on Deposit Guarantee Schemes [recast], COM (2010)368. Daviter, F. (2009) Schattschneider in Brussels: How Policy Conflict Reshaped the Biotechnology Agenda in the European Union, West European Politics 32(6): 1118-1139. Eising, R. (2004) Multi-level governance and business interests in the European Union, Governance. An International Journal of Policy, Administration and Institutions 17(2) 211-246. Entman, R. (1993) Framing: Toward Clarification of a Fractured Paradigm, Journal of Communication 43(4): 51-58. Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism, Princeton: Princeton University Press. Green Cowles, M., Caporaso, J. A., Risse-Kappen, T. (2001) Transforming Europe: Europeanization and Domestic Change, USA: Cornell University Press. Hall, P. A., and Soskice, D. (eds) (2001) Varieties of Capitalism. The Institutional Foundations of Comparative Advantage, New York: Oxford University Press. Klüver, H, Braun, C. and Beyers, J. (2014) <THIS ISSUE: PUBLISHER TO ADD /UPDATE DETAILS AT PROOF> Klüver, H., Mahoney, Ch., Opper, M. (2014) <THIS ISSUE: PUBLISHER TO ADD /UPDATE DETAILS AT PROOF> 21

Lijphart, A. (1999) Patterns of Democracy. Government Forms and Performance in Thirty- Six Countries, New Haven: Yale University Press. Majone, G. (1996) Regulating Europe, London: Routledge. Mansbridge, J. (1990) Beyond self-interest, Chicago, IL: The University of Chicago Press. Marks, G. (1993) Structural Policy and Multilevel Governance in the EC, in A. W. Cafruny and G.G. Rosenthal (eds), The State of the European Community, Vol 2: The Maastricht Debates and Beyond, Boulder Col: Harlow Longman, pp. 391-410. Moravcsik, A. (1998) The Choice for Europe. Social Purpose and State Power from Messina to Maastricht, London: UCL Press.. Olson, M. (1965) The Logic of Collective Action. Public goods and the theory of groups, Cambridge, MA: Harvard University Press. Ringe, N. (2005) Policy Preference Formation in Legislative Politics: Structures, Actors, and Focal Points, American Journal of Political Science 49(4): 731-745. De Vreese, C. (2005) News framing: Theory and typology. Information Design Journal & Document, Design13(1): 51 62. Schattschneider, E. E. (1960) The Semi-Sovereign People, New York: Holt, Rinehart and Winston. Schmitter, Philippe C. (1974) Still the Century of Corporatism?, The Review of Politics 36(1): 85-131. Schreier, M. (2012) Qualitative Content Analysis in Practice, Los Angeles: Sage. Stone Sweet, A., Sandholtz, W., Fligstein, N. (2001) The Institutionalization of Europe, Oxford, UK: Oxford University Press. Wallace, H. (2010): An Institutional Anatomy and Five Policy Modes, in: H. Wallace, M.A. Pollack, and A.R. Young (eds.): Policymaking in the European Union, Oxford: Oxford University Press, 69-104. 22

23

Table 1 Generic and specific frames in EU policymaking Frames WEEE RED AIFM DGS Total Generic institutional frames Implementation 5 0 1 1 7 Regulation 7 3 26 2 38 Harmonization 5 11 18 29 63 Administrative and economic 9 4 9 1 23 burdens Information and transparency 0 3 1 3 7 Specific institutional frame Mandatory targets 0 24 0 0 24 Generic policy frames Market integration 2 0 4 5 11 Employment 0 0 2 0 2 Consumer safety 0 0 6 37 43 Health 2 0 0 0 2 Specific policy frames Financial market stability 0 0 9 26 35 Environment 17 23 0 0 40 Climate 0 7 0 0 7 Energy 0 16 0 0 16 24

Table 2 The definition of the independent variables H Concept Variable definition Institutional context 1.1 EU level actors Dummy variable for EU level actors (reference category: Swedish actors) 1.2 EU member states Three dummy variables for German, British, and Dutch actors (reference category: Swedish actors) Policy context 2.1 Policy area 1=Environment, 0=Finance and Economy Strategic mobilization and framing 3.1 Political mobilization Number of policy documents in a policy debate 3.2 Policy position 0=actor has no clear position or is in favour of policy proposal, 1=actor opposes the proposed policy changes 3.3 Policy dimension: constituency interests 3.4 Policy dimension: Policy relevance 3.5 Policy dimension: normative implications 3.6 Policy dimension: remedy and costbenefit Control variables Policy scope Extent of policy change Type of actor 2=arguments referring to the interests of policy addressees and the EU public, 1=arguments referring to the interests of policy addressees or the EU public, 0=no such arguments 2=arguments referring to expert evidence and the urgency of the proposed measures, 1=arguments referring to expert evidence or the urgency of the proposed measures, 0=no such arguments 2=arguments referring to broad common goals and universal norms, 1=arguments referring to broad common goals or universal norms, 0=no such arguments 2=arguments referring to the distribution of costs and benefits and the social benefits of the proposed measures, 1=arguments referring to the distribution of costs and benefits or the social benefits of the proposed measures, 0=no such arguments Number of policy areas to which a policy document refers (ranging from 0 to 6) 1=major change or entirely new policy, 0=incremental or routine change Dummy variables for public interest groups, business interest groups, firms, institutions (reference category: EU and national state actors) 25

Table 3 Accounting for framing activities in EU policy debates: Poisson regressions Model 1 Model 2 Model 3 Model 4 Total number of No. of generic No. of generic No. of specific frames institutional frames policy frames policy frames German actors -0.438** -0.284-0.659-0.568 (0.206) (0.317) (0.494) (0.418) Dutch actors -0.0862 0.0405-0.631* -0.0193 (0.111) (0.170) (0.350) (0.206) British actors -0.111** -0.0908-0.235* -0.121 (0.0532) (0.0810) (0.136) (0.105) EU level actors -0.0851*** -0.0748* -0.0469-0.122** (0.0290) (0.0452) (0.0645) (0.0587) Policy area -1.795*** 0.375-5.425*** -1.806* (0.485) (0.733) (1.320) (0.972) Political mobilization 0.0280** -0.0423** 0.0687** 0.0524** (0.0123) (0.0185) (0.0320) (0.0247) Policy position -0.0365-0.0723 0.355-0.155 (0.139) (0.208) (0.299) (0.280) Dimension: policy relevance 0.304*** 0.402*** 0.303 0.184 (0.0922) (0.143) (0.250) (0.171) Dimension: normative justification 0.318*** 0.331** 0.341 0.385** (0.0866) (0.133) (0.212) (0.153) Dimension: constituency interests 0.286*** 0.197 0.448** 0.359** (0.0860) (0.133) (0.210) (0.156) Dimension: costs and benefits 0.221** -0.00368 0.266 0.414** (0.102) (0.171) (0.243) (0.170) Constant -1.704** 1.855* -5.421** -4.959*** (0.752) (1.102) (2.108) (1.510) LL -367.59-232.82-104.33-187.56 LR Chi2 114.91 72.13 109.71 62.02 P 0.000 0.000 0.000 0.000 Pseudo R2.1352.1341 0.3446 0.1419 Note: * significant at p = 0.1, ** significant at p = 0.05 significant at p = 0.01. Control variables (scope of policy, extent of policy change, and type of actor all have insignificant effects (not reported). 26